2 from __future__ import unicode_literals
6 from .common import InfoExtractor
13 class ARDIE(InfoExtractor):
14 _VALID_URL = r'^https?://(?:(?:www\.)?ardmediathek\.de|mediathek\.daserste\.de)/(?:.*/)(?P<video_id>[^/\?]+)(?:\?.*)?'
17 'url': 'http://www.ardmediathek.de/das-erste/guenther-jauch/edward-snowden-im-interview-held-oder-verraeter?documentId=19288786',
18 'file': '19288786.mp4',
19 'md5': '515bf47ce209fb3f5a61b7aad364634c',
21 'title': 'Edward Snowden im Interview - Held oder Verräter?',
22 'description': 'Edward Snowden hat alles aufs Spiel gesetzt, um die weltweite \xdcberwachung durch die Geheimdienste zu enttarnen. Nun stellt sich der ehemalige NSA-Mitarbeiter erstmals weltweit in einem TV-Interview den Fragen eines NDR-Journalisten. Die Sendung vom Sonntagabend.',
23 'thumbnail': 'http://www.ardmediathek.de/ard/servlet/contentblob/19/28/87/90/19288790/bild/2250037',
25 'skip': 'Blocked outside of Germany',
28 def _real_extract(self, url):
29 # determine video id from url
30 m = re.match(self._VALID_URL, url)
32 numid = re.search(r'documentId=([0-9]+)', url)
34 video_id = numid.group(1)
36 video_id = m.group('video_id')
38 webpage = self._download_webpage(url, video_id)
40 title = self._html_search_regex(
41 [r'<h1(?:\s+class="boxTopHeadline")?>(.*?)</h1>',
42 r'<h4 class="headline">(.*?)</h4>'],
44 description = self._html_search_meta(
45 'dcterms.abstract', webpage, 'description')
46 thumbnail = self._og_search_thumbnail(webpage)
50 for mo in re.finditer(
51 r'mediaCollection\.addMediaStream\((?P<media_type>\d+), (?P<quality>\d+), "(?P<rtmp_url>[^"]*)", "(?P<video_url>[^"]*)", "[^"]*"\)', webpage)]
53 if '"fsk"' in webpage:
54 raise ExtractorError('This video is only available after 20:00')
59 'quality': int(s['quality']),
62 format['protocol'] = 'rtmp'
63 format['url'] = s['rtmp_url']
64 format['playpath'] = s['video_url']
66 format['url'] = s['video_url']
68 quality_name = self._search_regex(
69 r'[,.]([a-zA-Z0-9_-]+),?\.mp4', format['url'],
70 'quality name', default='NA')
71 format['format_id'] = '%s-%s-%s-%s' % (
72 determine_ext(format['url']), quality_name, s['media_type'],
75 formats.append(format)
77 self._sort_formats(formats)
82 'description': description,
84 'thumbnail': thumbnail,